Reuse collected sub fields for objects of the same type by Meemaw · Pull Request #4501 · graphql-java/graphql-java

Meemaw · 2026-10-18T13:04:15Z

Problem

ExecutionStrategy.completeValueForObject runs FieldCollector.collectFields for every object in the result. For a list of N objects this repeats the same work N times: walking the selection set, evaluating @skip/@include, expanding fragments and building a fresh LinkedHashMap + MergedFields + MergedSelectionSet.

The result only depends on the merged field, the resolved object type, the variables and the fragments, and the last two are fixed for an execution. Async-profiler on ExecutionBenchmark showed field collection at ~16% of CPU and ~20% of allocated bytes.

Change

The collected sub fields are memoized per execution in ExecutionContext, keyed by MergedField and then by the resolved GraphQLObjectType. Every object of the same type under the same field reuses the same immutable MergedSelectionSet. Sub fields are shared too, so cache hits cascade down the tree.

Key correctness: MergedField.equals compares the class and the AST Field nodes (Field uses identity), so two merged fields are only equal when they come from exactly the same set of AST fields. Interface/union results are keyed per concrete object type.
Fallback to the old per-object collection when:
- incremental support (@defer) is enabled, because DeferredExecution instances are created per collection, or
- a custom ConditionalNodeDecision is in the GraphQLContext, because the callback is not guaranteed to be deterministic.
Memory: the cache is bounded by the number of distinct (field, object type) pairs in the operation and lives only as long as the ExecutionContext.
API surface: this adds a field to ExecutionContext plus one @Internal getter, getCollectedSubFields(). Happy to move it elsewhere if you prefer.

Benchmarks

Apple Silicon, JDK 21. Base vs this branch jars were run interleaved (alternating, 3 rounds × 2 forks each = 6 forks per side) to cancel out machine drift. Δ shows the 95% confidence interval.

Benchmark	Throughput Δ	Allocated bytes/op
`ExecutionBenchmark.benchmarkThroughput`	+14.7% ± 7.0%	1,073,041 → 903,689 (−15.8%)
`SimpleQueryBenchmark.benchMarkSimpleQueriesThroughput`	+10.2% ± 9.8%	1,933,273 → 1,510,351 (−21.9%)
`IntrospectionBenchmark` (large-schema-2)	+19.4% ± 9.0%	112.3M → 98.2M (−12.6%)¹

An earlier run on a quieter machine of the same change (before a cosmetic computeIfAbsent refactor) gave tighter intervals: ExecutionBenchmark +9.3% ± 0.8%, SimpleQuery +15.0% ± 1.1%, Introspection +18.4% ± 8.1%. An A/A run (base vs base) on that machine showed no significant difference, with a noise floor of about ±1.8% on ExecutionBenchmark.

¹ Introspection bytes/op vary between forks even for the same jar (96.9M vs 114.5M observed for base), so treat that row as indicative only.

Tests

New CollectedSubFieldsReuseTest covers:

objects of different concrete types in the same interface list each get their own type's fields
aliased fields with different sub selections are collected separately
fields merged across fragments keep all their sub selections
collection happens once per (field, object type)
a non-deterministic ConditionalNodeDecision is still asked for every object, so the cache is bypassed
the cache is bypassed when incremental support is enabled

To check the tests guard the fallbacks, I temporarily removed the bypass condition: the ConditionalNodeDecision test then returned wrong data and the incremental test failed. The full test suite passes, apart from the 14 locale-dependent StringValueParsingUnicodeTest failures that also occur on master with a non-English JVM locale (fixed separately in #4500).

🤖 Generated with Claude Code

@include

completeValueForObject collected the sub fields of the current field for every object in the result, so a list of N objects repeated the same selection set walk, @skip/@include evaluation and fragment expansion N times. The result only depends on the merged field and the object type, since variables and fragments are fixed for an execution, so it is now memoized per execution in ExecutionContext. The cache is bypassed when incremental support is enabled (deferred executions are created per collection) or when a custom ConditionalNodeDecision is present (it may not be deterministic). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

github-actions · 2026-10-30T20:24:40Z

Test Report

Test Results

Java Version	Total	Passed	Failed	Errors	Skipped
Java 11	-	-	-	-	-
Java 17	5968 (-85 🔴)	5911 (-85 🔴)	0 (±0)	0 (±0)	57 (±0)
Java 21	-	-	-	-	-
Java 25	5968 (-85 🔴)	5911 (-85 🔴)	0 (±0)	0 (±0)	57 (±0)
jcstress	32 (±0)	32 (±0)	0 (±0)	0 (±0)	0 (±0)
Total	11968 (-170 🔴)	11854 (-170 🔴)	0 (±0)	0 (±0)	114 (±0)

Code Coverage (Java 25)

Metric	Covered	Missed	Coverage	vs Master
Lines	30151	3117	90.6%	-0.1% 🔴
Branches	8846	1520	85.3%	-0.1% 🔴
Methods	8053	1206	87.0%	-0.2% 🔴

Changed Class Coverage (25 classes)

Class	Line	Branch	Method
g.a.FieldComplexityEnvironment	-7.7% 🔴	±0.0%	-14.3% 🔴
g.e.ExecutionStrategy	±0.0%	+0.5% 🟢	+0.1% 🟢
g.e.i.Instrumentation	+3.8% 🟢	±0.0%	+4.5% 🟢
g.l.AstPrinter	±0.0%	+0.4% 🟢	±0.0%
g.l.FloatValue $Builder	-7.1% 🔴	±0.0%	-9.1% 🔴
g.p.UnicodeUtil	-4.7% 🔴	±0.0%	±0.0%
g.r.SimpleListConnection	+0.4% 🟢	+0.3% 🟢	±0.0%
g.s.GraphQLAppliedDirective	-5.9% 🔴	±0.0%	-12.5% 🔴
g.s.GraphQLAppliedDirectiveArgument	-6.1% 🔴	±0.0%	-10.5% 🔴
g.s.GraphQLArgument	-1.6% 🔴	±0.0%	-3.1% 🔴
g.s.GraphQLEnumType	-1.2% 🔴	±0.0%	-2.7% 🔴
g.s.GraphQLEnumValueDefinition	-2.4% 🔴	±0.0%	-4.2% 🔴
g.s.GraphQLInputObjectField	-1.8% 🔴	±0.0%	-3.4% 🔴
g.s.GraphQLInputObjectType	-1.8% 🔴	±0.0%	-3.1% 🔴
g.s.GraphQLInterfaceType	-1.6% 🔴	±0.0%	-3.4% 🔴
g.s.GraphQLScalarType	-9.3% 🔴	±0.0%	-12.5% 🔴
g.s.GraphQLUnionType	-1.8% 🔴	±0.0%	-3.8% 🔴
g.s.i.SchemaGeneratorAppliedDirectiveHelper	±0.0%	-6.9% 🔴	±0.0%
g.s.i.SchemaGeneratorHelper	±0.0%	-0.7% 🔴	±0.0%
g.s.i.SchemaPrinter	±0.0%	-0.4% 🔴	±0.0%
g.s.i.TypeDefinitionRegistry	-0.6% 🔴	±0.0%	-2.5% 🔴
g.s.v.AppliedDirectiveArgumentsAreValid	-5.1% 🔴	-4.5% 🔴	±0.0%
g.s.v.TypesImplementInterfaces	-0.9% 🔴	-5.8% 🔴	±0.0%
g.u.Anonymizer $1	-4.8% 🔴	-6.3% 🔴	-3.3% 🔴
g.u.Anonymizer $2	-3.3% 🔴	-2.8% 🔴	±0.0%

FieldComplexityEnvironment — method details

Method	Line	Branch
getArguments	0.0% (-100.0% 🔴)	—

FloatValue.Builder — method details

Method	Line	Branch
ignoredChars	0.0% (-100.0% 🔴)	—

UnicodeUtil — method details

Method	Line	Branch
parseAndWriteUnicode	92.3% (-7.7% 🔴)	100.0%

GraphQLAppliedDirective — method details

Method	Line	Branch
getDefinition	0.0% (-100.0% 🔴)	—
copy	0.0% (-100.0% 🔴)	—

GraphQLAppliedDirectiveArgument — method details

Method	Line	Branch
getDefinition	0.0% (-100.0% 🔴)	—
copy	0.0% (-100.0% 🔴)	—

GraphQLArgument — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

GraphQLEnumType — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

GraphQLEnumValueDefinition — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

GraphQLInputObjectField — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

GraphQLInputObjectType — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

GraphQLInterfaceType — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

GraphQLScalarType — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—
withNewChildren	0.0% (-100.0% 🔴)	—
lambda$withNewChildren$0	0.0% (-100.0% 🔴)	—

GraphQLUnionType — method details

Method	Line	Branch
copy	0.0% (-100.0% 🔴)	—

SchemaGeneratorAppliedDirectiveHelper — method details

Method	Line	Branch
buildAppliedDirective	100.0%	66.7% (-33.3% 🔴)
buildDirectiveArg	100.0%	66.7% (-16.7% 🔴)
buildAppliedArg	80.0%	33.3% (-16.7% 🔴)

SchemaGeneratorHelper — method details

Method	Line	Branch
lambda$buildScalar$0	100.0%	50.0% (-50.0% 🔴)

SchemaPrinter — method details

Method	Line	Branch
lambda$scalarPrinter$1	100.0%	50.0% (-50.0% 🔴)

TypeDefinitionRegistry — method details

Method	Line	Branch
getType	0.0% (-100.0% 🔴)	—
getType	0.0% (-100.0% 🔴)	—

AppliedDirectiveArgumentsAreValid — method details

Method	Line	Branch
checkArgument	93.3%	85.7% (-7.1% 🔴)
isValidExternalValue	50.0% (-50.0% 🔴)	—

TypesImplementInterfaces — method details

Method	Line	Branch
lambda$checkFieldArgumentEquivalence$1	95.7% (-4.3% 🔴)	83.3% (-16.7% 🔴)

Anonymizer.1 — method details

Method	Line	Branch
visitGraphQLAppliedDirective	92.3% (-7.7% 🔴)	75.0% (-25.0% 🔴)
visitGraphQLScalarType	40.0% (-60.0% 🔴)	50.0% (-50.0% 🔴)
lambda$visitGraphQLScalarType$0	0.0% (-100.0% 🔴)	—

Anonymizer.2 — method details

Method	Line	Branch
visitGraphQLScalarType	40.0% (-60.0% 🔴)	50.0% (-50.0% 🔴)

Full HTML report: build artifact jacoco-html-report

Updated: 2026-10-30 20:28:46 UTC

Meemaw marked this pull request as ready for review October 18, 2026 17:45

Merge branch 'master' into perf/reuse-collected-sub-fields

8bfb212

Navigation Menu

Repository navigation

Reuse collected sub fields for objects of the same type - #4501

Reuse collected sub fields for objects of the same type#4501
Meemaw wants to merge 2 commits into
graphql-java:masterfrom
Meemaw:perf/reuse-collected-sub-fields

Meemaw commented Oct 18, 2026

Uh oh!

github-actions Bot commented Oct 30, 2026 •

edited

Loading

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

📋 Server Request & Proxy Logs

⚙️ Pengaturan Proxy Target

Conversation

Meemaw commented Oct 18, 2026

Problem

Change

Benchmarks

Tests

Uh oh!

github-actions Bot commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Test Report

Test Results

Code Coverage (Java 25)

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

github-actions Bot commented Oct 30, 2026 •

edited

Loading